Visual Trends
Detailed Data Breakdown
Type Cost
Showing 1 to 10 of 0 entries
Energy Usage Recommendations
Model Performance Evaluation (2024 Predictions)

This evaluation compares model predictions against actual data for 2024. Models were trained on historical data up to the start of 2024 and used to predict 2024 values.

DAILY Model Metrics
Mean Absolute Error
$4.0198
Root Mean Squared Error
$4.3032
Mean Absolute % Error
111.58%
R² Score
-6.4491
Mean Actual Value
$5.72
Mean Predicted Value
$1.95
Sample Size: 366 predictions
Mean Error: $-3.77
RMSE as % of Mean: 75.19%
WEEKLY Model Metrics
Mean Absolute Error
$7.7653
Root Mean Squared Error
$8.9596
Mean Absolute % Error
25.61%
R² Score
-1.1743
Mean Actual Value
$39.52
Mean Predicted Value
$33.00
Sample Size: 53 predictions
Mean Error: $-6.52
RMSE as % of Mean: 22.67%
MONTHLY Model Metrics
Mean Absolute Error
$19.9326
Root Mean Squared Error
$22.4822
Mean Absolute % Error
11.10%
R² Score
-2.7958
Mean Actual Value
$174.55
Mean Predicted Value
$155.61
Sample Size: 12 predictions
Mean Error: $-18.94
RMSE as % of Mean: 12.88%
About This Project

Motivation

Southern California experiences some of the highest electricity demand in the United States due to a combination of factors such as widespread air-conditioning use, a growing number of electric vehicles, and increasing residential and commercial energy consumption. During hot summer months, cooling loads drive peak demand in the late afternoon and evening, while electric vehicle charging and household activities further elevate nighttime consumption. These demand patterns often lead to periods of high electricity prices, even when consumers are unaware of the cost differences throughout the day.

At the same time, many energy-intensive activities—such as EV charging, running laundry, or operating large appliances—can be shifted to hours when electricity is cheaper. Identifying these “optimal usage windows” has the potential to reduce household energy bills, ease stress on the electric grid, and support more efficient use of renewable generation.

However, hourly electricity prices are not always publicly available for Southern California, and consumers rarely have access to clear or actionable guidance about when electricity is most affordable. This project addresses this gap by estimating hourly electricity costs using available demand, generation, and weather data, and by forecasting prices for the next day. With these predictions, the system provides users with intuitive recommendations about the best times to use electricity.

By helping consumers shift demand to lower-cost hours, the project supports both economic savings and grid reliability, while also encouraging more sustainable energy behavior in a region where electricity demand continues to rise.

Energy Usage Assistant
×
Assistant: Hi! I'm your energy usage assistant. I analyze predictions and historical data to help you save money. Ask me about the best times to use electricity, or I can automatically analyze today's data for you.